Re: About releasing vcpu when closing vcpu fd

[Anshul Makkar <anshul.makkar@profitbricks.com>]

Hi,

Currently as per the specs for cpu_hot(un)plug, ACPI GPE Block:  IO
ports 0xafe0-0xafe3  where each bit corresponds to each CPU.

Currently, EJ0 method in acpi-dsdt-cpu-hotplu.dsl doesn't do anything.

Method(CPEJ, 2, NotSerialized) {
        // _EJ0 method - eject callback
        Sleep(200)
    }

I want to implement a notification mechanism for CPU hotunplug just
like we have in memory hotunplug where in we write to particular IO
port and this read/write is caught in the memory-hotplug.c .

So, just want a suggestion as to whether I should expand the IO port
range from 0xafe0 to 0xafe4 (addition of 1 byte), where last byte is
for notification of EJ0 event.

Or if you have any other suggestion, please share.

Thanks
Anshul Makkar

On Fri, Jun 6, 2014 at 3:41 PM, Anshul Makkar
<anshul.makkar@profitbricks.com> wrote:
> Oh yes, sorry for the ambiguity.  I meant proposal to "park" unplugged vcpus.
>
> Thanks for the suggesting the practical approach.
>
> Anshul Makkar
>
> On Fri, Jun 6, 2014 at 3:36 PM, Gleb Natapov <gleb@minantech.com> wrote:
>> On Fri, Jun 06, 2014 at 03:02:59PM +0200, Anshul Makkar wrote:
>>> IIRC, Igor was of the opinion that  patch for vcpu deletion will be
>>> incomplete till its handled properly in kvm i.e vcpus are destroyed
>>> completely. http://comments.gmane.org/gmane.comp.emulators.kvm.devel/114347
>>> .
>>>
>>> So can the above proposal  where just vcpus can be  disabled and
>>> reused in qemu is an acceptable solution ?
>>>
>> If by "above proposal" you mean the proposal in the email you linked,
>> then no since it tries to destroy vcpu, but does it incorrectly. If you
>> mean proposal to "park" unplugged vcpu, so that guest will not be able
>> to use it, then yes, it is pragmatic path forward.
>>
>>
>>> Thanks
>>> Anshul Makkar
>>>
>>> On Thu, May 29, 2014 at 10:12 AM, Gleb Natapov <gleb@kernel.org> wrote:
>>> > On Thu, May 29, 2014 at 01:40:08PM +0800, Gu Zheng wrote:
>>> >> >> There was a patch(from Chen Fan, last august) about releasing vcpu when
>>> >> >> closing vcpu fd <http://www.spinics.net/lists/kvm/msg95701.html>, but
>>> >> >> your comment said "Attempt where made to make it possible to destroy
>>> >> >> individual vcpus separately from destroying VM before, but they were
>>> >> >> unsuccessful thus far."
>>> >> >> So what is the pain here? If we want to achieve the goal, what should we do?
>>> >> >> Looking forward to your further comments.:)
>>> >> >>
>>> >> > CPU array is accessed locklessly in a lot of places, so it will have to be RCUified.
>>> >> > There was attempt to do so 2 year or so ago, but it didn't go anyware. Adding locks is
>>> >> > to big a price to pay for ability to free a little bit of memory by destroying vcpu.
>>> >>
>>> >> Yes, it's a pain here. But if we want to implement "vcpu hot-remove", this must be
>>> >> fixed sooner or later.
>>> > Why?  "vcpu hot-remove" already works (or at least worked in the past
>>> > for some value of "work").  No need to destroy vcpu completely, just
>>> > park it and tell a guest not to use it via ACPI hot unplug event.
>>> >
>>> >> And any guys working on kvm "vcpu hot-remove" now?
>>> >>
>>> >> > An
>>> >> > alternative may be to make sure that stopped vcpu takes as little memory as possible.
>>> >>
>>> >> Yeah. But if we add a new vcpu with the old id that we stopped before, it will fail.
>>> >>
>>> > No need to create vcpu again, just unpark it and notify a guest via ACPI hot plug event that
>>> > vcpu can be used now.
>>> >
>>> > --
>>> >                         Gleb.
>>> > --
>>> > To unsubscribe from this list: send the line "unsubscribe kvm" in
>>> > the body of a message to majordomo@vger.kernel.org
>>> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>
>> --
>>                         Gleb.